Date of Award

Fall 2016

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

First Advisor

Schneider, Susan C.

Second Advisor

Yaz, Edwin E.

Third Advisor

Jeong, Chung Seop

Fourth Advisor

Heinen, James

Fifth Advisor

Richie, James

Abstract

This dissertation addresses the dynamic control of nonlinear systems with finite energy noise in the state and measurement equations. Regional eigenvalue assignment (REA) is used to ensure that the state estimate error is driven to zero significantly faster than the state itself. Moreover, the controller is designed for the resulting closed loop system to achieve any one of a set of general performance criteria (GPC). The nonlinear model is assumed to have a Lipschitz nonlinearity both in the state and measurement equations. By using the norm bound of the nonlinearity, the controller is designed to be robust against all nonlinearities satisfying the norm-bound. A Luenberger-type nonlinear observer is used to estimate the system state, which is not directly measurable. The choice of the eigenvalue locations for the linear part of the system is based on the transient response specifications and the separation of the controller dynamics from the observer dynamics. Furthermore, the GPC are incorporated to achieve performance requirements such as H2, H∞, etc. The advantage of using GPC is it allows the designer flexibility in choosing a performance objective to tune the system. The design problem introduced in this dissertation uses various mathematical techniques to derive LMI conditions for the controller and observer design using REA, GPC, and the bounds on the Lipschitz nonlinearities. All work will be demonstrated in both continuous- and discrete-time. Illustrative examples in both time domains are given to demonstrate the proposed design procedure. Multiple numerical approaches are also presented and compared in simulations for ease of use, applicability, and conservatism.

Share

COinS